Sebastian Maerkl joined the California Institute of Technology in 2001 looking to pursue his doctoral research in the field of proteomics. He picked the eukaryotic transcription factors belonging to the basic-helix-loop-helix (bHLH) family for his initial studies. “They are about the third largest family of transcription factors, involved in a wide variety of cellular functions and they form very interesting networks,” says Maerkl. “The bHLH transcription factors can form homo as well as heterodimers with their protein partners, leading to highly complex structures and, being transcription factors, they also bind to DNA.”

Figure 1. A microfluidic device for performing 2400 parallel reactions to measure molecular interactions using mechanically induced trapping of molecular interactions (MITOMI). The control lines have been filled with dyes for visualization.

To study these transcription factors he started out using a standard protein array platform in which targets are exposed to proteins bound to a surface. “I soon realized that with transcription factors it was very difficult to measure transient interactions using this method,” says Maerkl. The stringent washing requirements that were a part of the experimental protocol were the biggest limitation.

“The bHLH transcription factors that we studied have an off rate of about 0.2 per second, which yields a half life of about 3 seconds,” says Maerkl. “In other words, when you let your target DNA bind to the transcription factor, and as you wash after about 3 seconds, half of the material bound to the surface has disappeared. So when you use the standard microarray protocol with 30 minute wash steps only the top binders yield a signal.”

He then met Stephen Quake, a professor in the department of bioengineering at Stanford University who, just the previous year, had published a paper (Unger et al., Science 288:113-118 (2000)) describing the creation of a microfluidic device consisting of flow channels with pneumatically activated valves and pumps. Intrigued by the possibility of using such a microfluidic, lab-on-a-chip type device for his study, Maerkl started working with Quake, and spent the first two years of his graduate career refining the multi-layer soft lithography technique that Quake had developed.

In 2002 he was part of the team that published another paper (Thorsen et al., Science 298:581-584 ( 2002)) describing the use of microfluidic large-scale integration (MLSI) to develop a microfluidic array with 1000 individual picoliter-scale reaction chambers that could be tailored for various applications. “MLSI allows you to create multiple layers of flow structures which in combination form functional elements that help perform complicated fluid manipulations,” says Maerkl.

Maerkl then succeeded in using the MLSI-based high-throughput microfluidic platform for detecting the low-affinity transient interactions seen with the bHLH proteins. Earlier this year Maerkl published his work (Maerkl and Quake, Science 315:233-237 (2007)) describing what he refers to as MITOMI, which stands for mechanically-induced trapping of molecular interactions. “It’s a very simple idea where we use one of our actionable membranes not to shunt flow in a microfluidic device but to trap molecules on the surface to detect interactions,” he says.

Figure 2. Magnified section of the central grid of the device shown in Figure 1. The individual dumbbell shaped unit cells are clearly discernible. Each unit cell is controlled by a set of two valves (orange and green lines), and contains a button membrane (blue line) for performing MITOMI.

Maerkl has used MITOMI to gather over 41,000 individual data points, with more than 17 different microfluidic devices and measuring over 464 target DNA sequences to map the binding energy landscapes of the bHLH proteins. The binding energy topographies were then used to predict their in vivo functions and the structural elements that determine their binding specificities.

The advantage of the technique is its scalability, says Maerkl. He feels that routinely used optical methods like surface plasmon resonance (SPR) and total internal reflection fluorescence (TIRF) microscopy, though good for studying kinetics and affinity constants, do not offer the parallelization achievable with MITOMI. “Currently the device offers the ability to perform 2400 measurements simultaneously. The logical next step would be to combine MITOMI with a standard DNA microarray to enable the measurement of binding energies of thousands of parallel sequences of transcription factors,” says Maerkl. The use of microfluidics also eliminates the washing steps that conventional array platforms demand. “Instead of washing you basically squeeze away the liquid and protect that region of the device so that there is no further loss of material from the surface.”

As Maerkl gets ready to graduate this year, members of the Quake lab are looking to expand the technique’s versatility by using it to study other types of molecular interactions, such as protein-protein, protein-DNA, protein-RNA, protein-small molecule. “MITOMI should be more broadly applicable and be able to yield more data in the future,” says Maerkl.

Making Microfluidics More Accessible